Taalportaal - the digital language portal

Dutch
Frisian
Saterfrisian
Afrikaans

Show all

Complex segments as single units

quickinfo

Consonant clusters of /s/ and a voiceless plosive — so-called 'complex segments' — act as single phonological units, for which several arguments can be adduced. They are the topic of this section.

Consonant clusters of /s/ and a voiceless plosive act as single phonological units.

A first peculiarity of these clusters is that they can occur both word-initially and word-finally, with the obstruents appearing in either order (see clusters of /s/ + voiceless plosive in word-initial position and clusters of /s/ + voiceless plosive in word-final position). They do not seem to conform to the Sonority Sequencing Constraint, in marked contrast with clusters of an obstruent and a sonorant consonant, which are each other's mirror image in word-initial and word-final position.

There are two more pieces of evidence that these clusters act as units. In the first place, it is observed in Hoekstra (1985:70-71) that degemination (see degemination) not only affects double consonants, but also double /s/ + plosive clusters, which only show up in compounds and phrases, as shown in (1):

list#stik	[st]	piece of a frame
fisk#skaal	[sk]	fish dish
de gasp spanne	[sp]	to tighten the buckle
op 'e kast stean	[st]	to stand on the cupboard
de wask skjin	[sk]	the laundry clean
Geps psalmboek	[ps]	Gep's psalmbook
oer de nije fyts tsiere	[ts]	to quarrel about the new bicycle
is dat faaks ksenon	[ks]	does this happen to be xenon?

For degemination to occur, these sequences must have the status of a single unit at some level of representation.

In the second place, /s/ + plosive clusters cannot be split by schwa-epenthesis, whereas a branching onset, of an obstruent and a liquid, can (see schwa insertion in onset clusters). So, trien /triən/ tear can be pronounced as [təri.ən], but tsien /tsiən/ ten or stien /stiən/ stone not as [*təsi.ən] or [*səti.ən]. Insertion would disrupt the unit which the sequence of /s/ + plosive is.

The /s/+ plosive clusters are not complex segments at the level of underlying representation, i.e. they are the result of phonological processing, for which several arguments can be adduced.

One argument of formal economy is that if complex segments arise through processing, the phoneme inventory of Frisian can be kept much simpler than when they are assigned phonemic status.

Another argument is that these sequences also show up in morphologically derived structures. There are three cases at hand. First, the possessive suffix -s ( /-s/) can be attached to a proper name ending in /-p/, /-t/, or /-k/ in order to mark the possessor of the object denoted by the following noun (see the possessive suffix -s), as in Geps /ɡɛp+s/ Gep's, Juks /jøk+s/ Juk's, and Reits /rajt+s/ Reit's. Second, the suffix -s ( /-s/) can also be attached to an adjective ending in /-p/, /-t/, or /-k/ to form a 'partitive genitive' (see partitive construction), as in wat knaps /knap+s/ something smart, wat wyts /vit+s/ something white, and wat bryks /brik+s/ something crooked. Third, the third person singular present tense verbal suffix -t ( /-t/) can be attached to a verb stem (of the first weak or strong/irregular class, see paradigm of class I) ending in /-s/, as in hy/hja past /pɔs+t/ [pɔst]hjir he/she belongs here.

/s/ + plosive clusters can also result from phonological processes which 'act upon' obstruent sequences derived by morphology. When a proper name ends in voiced /-b/ or /-d/, these show up as [p] or [t] when the possessive suffix is added, as in Sybs /sib+s/ [sips] Syb's or Tseards /tsɪəd+s/ [tsɪəts] Tseard's. The same happens when an adjective ending in /-b/ or /-d/ is suffixed with partitive /-s/, as in wat sibs /sɪb+s/ [sɪps] something familiar and wat reads /rɪəd+s/ [rɪəts] something read.

Verbal morphology also gives rise to the clusters at hand. When a verbal stem ends in /-z/, the latter turns into [s] upon suffixation with /-t/ (3rd person singular, present tense), as in hy/hja gniist /ɡni:z+t/ [ɡni:st] he/she smirks/sneers. The suffix -st (2nd person singular, present tense) may have the same effect, be it in an indirect way. There are two instances to be considered. First, when a verbal stem ending in /-s/ is suffixed with -st, as in do skrast you strike out, the resulting form, /skrɔs+st/, undergoes degemination (see degemination): [skrɔst]. Second, stem-final /-z/ turns into [s] before the suffix -st; with subsequent degemination this yields a form with final [-st], as in do gniist you smirk/sneer /ɡni:z+st→ ɡni:sst→ gni:st/. These are all cases of obstruent devoicing before a word-final (voiceless) obstruent.

Take a verb like lige /li:ɣ+ə/ to tell a lie. When /li:ɣ/ is suffixed with -st, stem-final /-ɣ/ is turned into [-x], resulting in [lixst]. The marked fricative sequence [-xs-] may be repaired by the dissimilatory hardening of /x/ to /k/ (see dissimilation), resulting in [likst] (for the shortening of /-i:-/ to [-i-], see overview of the contexts of Vowel Shortening). Take also the noun nacht /naxt/ night, from which the temporal adverb nachts /naxt+s/ at night derives. Deletion of /t/ turns nachts into /naxs/, the final cluster of which may undergo dissimilation: [naks] (see dissimilation).

All these /s/ + plosive or plosive + /s/ clusters, whether derived morphologically or phonologically, act as units. First, they can participate in degemination, as in the examples in (2) (cf. those in (1)):

Degemination of derived /s/ + plosive clusters

Geps psalmboek

[ps]

Gep's psalmbook

wêr is Juks ksylofoan keard

[ks]

where has Juk's xylophone got to?

nachts

[naks]

ksylofoan spylje

[ks]

to play the xylophone at night

ik wit net, hoe't Jaaps psyche wurket

[ps]

I do not know how Jaap's psyche works

ik kaam wat wyts tsjin

[ts]

I came across something white

oer wat reads tsiere

[ts]

to quarrel about something red

dat past stilistysk

[st]

net

that does not fit stylistically

hy blaast stevich

[st]

he blows firmly

Second, they cannot be split by epenthesis: Geps [*ɡɛpəs], wat wyts [*vitəs], wat reads [*rɪətəs], hy/hja past [*pɔsət], hy/hja blaast [*bla:sət], nachts [*nakəs]. The assumption that all /s/ + plosive complex segments are the result of phonological processing, hence that they are not present as such in underlying representation, enables one to arrive at a generalized analysis.

extra

The fact that the verb forms [pɔsət] and [bla:sət] do not strike us as too bad may be ascribed to the fact that the 3rd person singular of verbs of the (unmarked) second weak class ends in [-ət] (see paradigm of class II).

Diminutive formation provides us with an argument that noun-final /-st/ is a sequence of two separate segments in underlying representation. Nouns ending in /-ft/ and /-st/, like skoft /skoft/ while, time and kast /kɔst/ cupboard, have two diminutive forms, viz. skoftsje /skoft+tsjə/ [skoftsjə] and kastje /kɔst+tsjə/ [kɔsjə], and skoftke /skof+kə/ [skofkə] and kastke /kɔs+kə/ [kɔskə]. The forms ending in [-jə] are found in the eastern part of Fryslân, those ending in [-kə] in the western part of Fryslân (see phonological adjustments). The forms skoftke and kastke arise through deletion of (extrasyllabic) stem-final /-t/, after which the diminutive allomorph -ke ( /-kə/) is selected on the basis of /-f/ or /-s/, which have become stem-final (cf. slof /slof/ slipper, mule --> slofke [slofkə] little slipper, mule and tas /tɔs/ bag --> taske [tɔskə] little bag). All this speaks in favour of an analysis in which stem-final /-st/ is a sequence of two separate segments in underlying representation, which is turned into a complex segment through phonological processing. Apart from the objections to underlying complex segments raised above, in this particular case they would force us to assume two kinds of /t/-deletion, one for the complex segment /-st/ and one for the sequence /-ft/, which would considerably and unnecessarily complicate the grammar.

Especially in the southern part of Fryslân, the initial /t/ of the sequence /tsj-/ is prone to deletion. Words like tsjerke /tsjɛrkə/ church, tsjettel /tsjɛtəl/ kettle, and tsjuster /tsjøstər/ dark; darkness are often pronounced there as [sjɛrkə], [sjɛtl̩], and [sjøstr̩].

extra

In general, this is not reflected in the spelling, though tsjoele and sjoele shovel occur side by side.

This deletion is best accounted for if /ts-/ is analyzed as a cluster of two separate consonants.

The occurrence of syllabic sonorant consonants provides a purely phonological argument that word-final /s/ + plosive clusters consist of two separate segments in underlying representation. The final sequences in clusters of /s + plosive in word-final position are, among others, found in nouns: gasp /gɔsp/ buckle, clasp, meeps /me:ps/ wasp, least /lɪəst/ last; footmark, guts /gøts/ gouge, fisk /fɪsk/ fish, and foks /foks/ fox. When a suffix of the form schwa + sonorant consonant is attached to these stems, as in the plurals gaspen /gɔsp+ən/ buckles, clasps, meepsen /me:ps+ən/ wasps, leasten /lɪəst+ən/ lasts; footmarks, gutsen /gøts+ən/ gouges, fisk /fɪsk+ən/ fishes, and foks /foks+ən/ foxes, these can be realized as /(ɡɔs)(pm̩)/, /(me:p)(sn̩)/, /(lɪəs)(tn̩)/, /(ɡøt)(sn̩)/, /(fɪs)(kŋ̩)/, and /(fok)(sn̩)/. The syllabic nasal provides strong evidence that /s/, /p/, /t/, and /k/ are in the onset of the final syllable, since a syllable headed by a (sonorant) consonant must have an onset (see the onset condition). This, in its turn, means that the stem-final clusters in the above examples are best analyzed as two separate consonants, both of which are syllabified independently from each other.

References

Hoekstra, Jarich1985t-deletion before suffix-initial st in modern West FrisianNOWELE : North-Western European language evolution563-76

⇑ Back to top

About this article

Author(s): Willem Visser

Category: Frisian Phonology

Keywords: complex segmentbehaviour as a single unit

Version: 1.0, December 2015

Version history:

version	editor(s)	date	remarks
1.0	Willem Visser	December 2015

printreport errorcite

Jump to:

introduction
references
related articles